Identification of Electrical Appliances Using Their Virtual Description and Data Selection for Non-Intrusive Load Monitoring

The proper pattern of electric energy management on the part of consumers is a key element of the system enabling its effective use. This pattern can be developed by providing consumers with information obtained from non-invasive load monitoring. Improvement electric energy management can be achieved by turning off unnecessary appliances and by shifting some of them to off-peak hours. A rational benefit for consumers is the reduction of electricity consumption costs. In this paper, energy disaggregation was abandoned by introducing virtual electrical devices and by using optimal selection of explanatory data obtained from high-frequency measurements. The concept of virtuality results from the simultaneous operation of many physical devices, and especially for which it is difficult to distinguish the state of load, e.g., small, medium or large. The explanatory data was selected using the integral information capacity indicator of individual features. The determination of this indicator is based on the correlations between the explanatory variables, the explained variable and the mutual correlations between the explanatory variables. The parameters selected in this way constitute input data for various artificial neural network architectures identifying the electrical appliances. The obtained research results are very promising and confirm the validity of the approach used.